// CHECK:           %[[VAL_14:.*]] = "tf.Abs"(%[[VAL_1]]) : (tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK:           %[[VAL_15:.*]] = "tf.Div"(%[[VAL_13]], %[[VAL_14]]) : (tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK:           %[[VAL_16:.*]] = "tf.Select"(%[[VAL_6]], %[[VAL_7]], %[[VAL_15]]) : (tensor<2x3xi1>, tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>

    %0 = stablehlo.convert %arg0 : (tensor<1x2xf32>) -> tensor<1x2xi32>
    return %0 : tensor<1x2xi32>
  }
// CHECK: @uniform_quantize_0
  func.func private @uniform_quantize_1(%arg0: tensor<1x3xf32>, %arg1: tensor<1x1xf32>, %arg2: tensor<1x1xi8>) -> tensor<1x3xi8> {
    %0 = stablehlo.convert %arg0 : (tensor<1x3xf32>) -> tensor<1x3xi8>

    %9 = "tf.Const"() {value = dense<-1> : tensor<1xi32>} : () -> tensor<1xi32>
    %10 = "tf.Const"() {value = dense<[[0, 0], [3, 3], [3, 3], [0, 0]]> : tensor<4x2xi32>} : () -> tensor<4x2xi32>
    %11 = "tf.Pad"(%arg0, %10) : (tensor<2x224x224x3xf32>, tensor<4x2xi32>) -> tensor<2x230x230x3xf32>
    %12 = "tf.Cast"(%arg1) {Truncate = false} : (tensor<2x1xf32>) -> tensor<2x1xi64>

    %17 = "tf.Mul"(%16, %cst_0) : (tensor<2x2xf32>, tensor<f32>) -> tensor<2x2xf32>
    %18 = "tf.AddV2"(%17, %cst) : (tensor<2x2xf32>, tensor<f32>) -> tensor<2x2xf32>
    %19 = "tf.Floor"(%18) : (tensor<2x2xf32>) -> tensor<2x2xf32>
    %20 = "tf.ClipByValue"(%19, %cst_5, %cst_6) : (tensor<2x2xf32>, tensor<f32>, tensor<f32>) -> tensor<2x2xf32>

  %s0 = "tf.Const"() {value = dense<[4, 1]> : tensor<2xi32>} : () -> tensor<2xi32>
  %s1 = "tf.Const"() {value = dense<[4, 4]> : tensor<2xi32>} : () -> tensor<2xi32>
  // expected-error @+1 {{failed to verify that result 0 is 1-D}}
  %r0, %r1 = "tf.BroadcastGradientArgs"(%s0, %s1) : (tensor<2xi32>, tensor<2xi32>) -> (tensor<2x2xi32>, tensor<0xi32>)
  func.return %r0, %r1 : tensor<2x2xi32>, tensor<0xi32>
}

// -----

  // CHECK:  %5 = "tfl.arg_max"(%arg0, %cst) : (tensor<1x32x1xf32>, tensor<1xi32>) -> tensor<1x1xi32>
  // CHECK:  return %4, %5 : tensor<1x1xf32>, tensor<1x1xi32>
}

// CHECK-LABEL:   func @convert_pytorch_argmax
func.func @convert_pytorch_argmax(%arg0: tensor<1x9xi32>) -> tensor<1xi32> {
  %0 = mhlo.constant dense<0> : tensor<i32>

  // CHECK-LABEL: @xla_svd
  func.func @xla_svd(%arg0: tensor<1x1xf32>) -> (tensor<1xf32>, tensor<1x1xf32>, tensor<1x1xf32>) {
    // CHECK-NOT: XlaSvd
    %s, %u, %v = "tf.XlaSvd"(%arg0) {max_iter = 1, epsilon = 1.0E-09 : f32, precision_config = ""} : (tensor<1x1xf32>) -> (tensor<1xf32>, tensor<1x1xf32>, tensor<1x1xf32>)
    func.return %s, %u, %v : tensor<1xf32>, tensor<1x1xf32>, tensor<1x1xf32>
  }

  func.func @identity(%arg0: f32) -> f32 {

    // CHECK:      %[[ONE:.*]] = "tf.Const"() <{value = dense<1> : tensor<i32>}>
    // CHECK:      %[[RES_READ_VAL:[0-9]*]] = "tf.ReadVariableOp"
    // CHECK-SAME: (tensor<*x!tf_type.resource<tensor<2x8xi32>>>) -> tensor<2x8xi32>
    // CHECK:      "tf.AddV2"(%[[RES_READ_VAL]], %[[ONE]])
    // CHECK-SAME: (tensor<2x8xi32>, tensor<i32>) -> tensor<2x8xi32>
    // CHECK:      "tf.AssignVariableOp"

}
func.func @_func(%arg0: tensor<2x4xf32>, %arg1: tensor<4x2xf32>) -> tensor<2x2xf32> {
  %0 = "tf.MatMul"(%arg0, %arg1) {_XlaSharding = "\08\03\1A\02\02\01\22\02\00\01"} : (tensor<2x4xf32>, tensor<4x2xf32>) -> tensor<2x2xf32>
  %1 = "tf.Identity"(%0) : (tensor<2x2xf32>) -> tensor<2x2xf32>
  return %1 : tensor<2x2xf32>
}

// -----
// The following op sharding is used in the following test case:

  %5 = "tf.Identity"(%4) {device = ""} : (tensor<2x4xf32>) -> tensor<2x4xf32>
  %6 = "tf.Identity"(%5) {device = ""} : (tensor<2x4xf32>) -> tensor<2x4xf32>
  func.return %6 : tensor<2x4xf32>

  // CHECK-LABEL: QuantDequantTranspose
  // CHECK-DAG: %[[CST:.*]] = "tf.Const"() <{value = dense<[1, 0]> : tensor<2xi32>}> : () -> tensor<?xi32>